Yesterday, the grammar checking service Grammarly announced that it will be launching GrammarlyGO, an AI writing tool that will be baked directly into their product.
Rahul Roy-Chowdhury, Grammarly’s global head of product, has said that the service will be based on GPT-3 large language model, putting it slightly behind Chat-GPT, which uses GPT 3.5.
According to Roy-Chowdhury, the tool will be able to help users rewrite content, including for tone, clarity and length, compose new text and help with brainstorming ideas.
Grammarly aims to launch the new tool in April, when it will be available to all customers, including free users. It will be enabled by default, though Grammarly for Education customers will have the option to disable it.
In late November 2022, OpenAI announced the public launch of ChatGPT. The product became an immediate hit, reaching over 100 million users in just two months.
However, the sudden popularity and efficacy of AI sent shockwaves through academia as students were able to generate assignments. This raised many questions schools should respond to AI, with many blocking access to ChatGPT on their networks.
For those already worried about the impact of AI writing on academia, this Grammarly decision is likely a new reason to be concerned, as it runs the risk of legitimizing plagiarism in the classroom.
How Grammarly Could Legitimize Plagiarism
To be clear, GrammarlyGo appears to be a very powerful and impressive tool. The demos provided by Grammarly showcase it crafting an outline for a project, composing emails with a specific tone, and doing powerful revisions of existing text.
However, instructors are rightfully concerned that Grammarly could, intentionally or unintentionally, be seen as legitimizing AI-powered plagiarism.
This is because Grammarly has marketed itself heavily as a tool for students to use on class assignments. They offer accounts both to individual students and to institutions, and target the education market aggressively.
As the company has expanded past basic spelling and grammar checking, it’s been a powerful and public anti-plagiarism force. Two of its most prominent features have been a plagiarism checker and a citation generator, which it added in September 2022.
Students trust and rely on Grammarly to help prevent plagiarism issues. However, now the company has made a left turn into making plagiarism much easier.
Though it’s unclear exactly how Grammarly will present this feature, it’s pretty clear that the company is leaning into the text generation features, at least when it comes to emails and other daily tasks. Whether it will warn students about the dangers of using AI-generated text for an assignment is unknown.
However, without (or even with) such a warning, many students will likely use it accordingly. The reason is not just because of greater accessibility, but because of Grammarly’s reputation for helping students prevent plagiarism issues.
This is how GrammarlyGO could be a major shift in the landscape of AI-generated plagiarism.
How it Changes the Playing Field
From a technology standpoint, very little changes with Grammarly’s announcement. Their product uses an older version of the large language model that powers ChatGPT. With ChatGPT being widely available for months, students have largely the same level of access that they had before GrammarlyGO.
Likewise, detection of AI-produced text doesn’t change. There are many tools and approaches available for detecting AI-generated text. Those tools will likely work just as well on GrammarlyGO as they would text from ChatGPT directly.
What GrammarlyGO does change is both the convenience of using an AI and, as discussed above, the legitimacy of it. Grammarly’s reputation as an anti-plagiarism tool gives AI a sense of credibility that schools have been working hard to dispel.
This has to be incredibly frustrating for schools, which have worked hard to caution students against the use of AI, only to have those efforts undermined by celebrity student athletes and now Grammarly.
Grammarly could, theoretically, include strong language in their product and present the tool in a way that is compatible with the rules schools have around the use of AI. However, that is almost impossible as many schools are currently trying to figure out what exactly those rules are.
In short, Grammarly has virtually ensured that a lot more students will experiment with AI writing in the classroom, with at least some not understanding that what they are doing is wrong because they trust the Grammarly name.
Why Grammarly Made This Move
All this begs a simple question: Why did Grammarly make this move, and why is it leaning so hard on text generation?
The likely reason was laid out in an article by Fast Company. Many users, in particular students, had discovered that ChatGPT could replace many of the core functions of Grammarly and some had cancelled their subscriptions, with one user calling it “pointless”.
In short, Grammarly was getting left behind and integrating AI into its own product is a means to bridge that gap.
And, to be clear, there are many ethical uses of a tool like this. For example, Grammarly demonstrates how it can be used to speed up email replies or other mundane tasks where authorship isn’t crucial. There are spaces where GrammarlyGO could be both useful and ethical. Grammarly is bolstering this by making sure it’s available pretty much anywhere one writes.
However, the dangers of abuse are also very real. Though schools have not responded to GrammarlyGO yet, I can’t imagine they are very happy about it given the potential for abuse.
In creative and academic spaces, GrammarlyGO could land human authors in a great deal of trouble. Grammarly needs to not only warn against that, but ensure that GrammarlyGO can’t be trivially used for those purposes. However, that doesn’t seem to be the direction that they are heading at this time.
As such, it’s only a matter of time before Grammarly goes from being a service that helps students paraphrase, cite and write better to one that gets them before disciplinary boards.
If it wasn’t clear before, it’s clear now: AI writing is here to stay.
With large tech companies like Google, Microsoft, Meta and Apple all investing heavily into AI, that was going to be true regardless of what Grammarly did.
However, by introducing AI writing into a tool that, historically, has been a tool for preventing plagiarism issues, gives it an unearned sense of legitimacy in the classroom.
While it may not change much in a practical sense, GrammarlyGO could be a powerful tool for changing the perception of AI, especially among students.
In the long run, that may be more impactful than any technological development in the space.
After all, the impact of a new technology isn’t just about what the technology can do, it’s about how it is used and where the ethical boundaries around it are drawn.